Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

libcusparse.so.11 not defined in file libcusparse.so.11 with link time reference #90692

Closed
ZhiyuanChen opened this issue Dec 12, 2022 · 10 comments
Assignees
Labels
high priority module: binaries Anything related to official binaries that we release to users module: cuda Related to torch.cuda, and CUDA support in general module: regression It used to work, and now it doesn't module: third_party triage review

Comments

@ZhiyuanChen
Copy link
Contributor

ZhiyuanChen commented Dec 12, 2022

🐛 Describe the bug

import torch
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
  File "/opt/conda/lib/python3.9/site-packages/torch/__init__.py", line 192, in <module>
    from torch._C import *  # noqa: F403
ImportError: /opt/conda/lib/python3.9/site-packages/torch/lib/libtorch_cuda_cu.so: symbol cusparseSpSM_analysis, version libcusparse.so.11 not defined in file libcusparse.so.11 with link time reference

Versions

PyTorch version: N/A
Is debug build: N/A
CUDA used to build PyTorch: N/A
ROCM used to build PyTorch: N/A

OS: CentOS Linux release 7.6.1810 (Core) (x86_64)
GCC version: (Anaconda gcc) 11.2.0
Clang version: 3.4.2 (tags/RELEASE_34/dot2-final)
CMake version: version 3.22.1
Libc version: glibc-2.17

Python version: 3.9.13 (main, Aug 25 2022, 23:26:10) [GCC 11.2.0] (64-bit runtime)
Python platform: Linux-3.10.0-1160.36.2.el7.x86_64-x86_64-with-glibc2.17
Is CUDA available: N/A
CUDA runtime version: 11.7.99
CUDA_MODULE_LOADING set to: N/A
GPU models and configuration: Could not collect
Nvidia driver version: Could not collect
cuDNN version: Could not collect
HIP runtime version: N/A
MIOpen runtime version: N/A
Is XNNPACK available: N/A

Versions of relevant libraries:
[pip3] mypy-extensions==0.4.3
[pip3] numpy==1.21.5
[pip3] numpydoc==1.4.0
[pip3] torch==1.13.0
[pip3] torchaudio==0.13.0
[pip3] torchvision==0.14.0
[conda] blas 1.0 mkl https://repo.anaconda.com/pkgs/main
[conda] ffmpeg 4.3 hf484d3e_0 pytorch
[conda] mkl 2021.4.0 h06a4308_640 https://repo.anaconda.com/pkgs/main
[conda] mkl-service 2.4.0 py39h7f8727e_0 https://repo.anaconda.com/pkgs/main
[conda] mkl_fft 1.3.1 py39hd3c417c_0 https://repo.anaconda.com/pkgs/main
[conda] mkl_random 1.2.2 py39h51133e4_0 https://repo.anaconda.com/pkgs/main
[conda] numpy 1.21.5 py39h6c91a56_3 https://repo.anaconda.com/pkgs/main
[conda] numpy-base 1.21.5 py39ha15fc14_3 https://repo.anaconda.com/pkgs/main
[conda] numpydoc 1.4.0 py39h06a4308_0 https://repo.anaconda.com/pkgs/main
[conda] pytorch 1.13.0 py3.9_cuda11.7_cudnn8.5.0_0 pytorch
[conda] pytorch-cuda 11.7 h67b0de4_0 pytorch
[conda] pytorch-mutex 1.0 cuda pytorch
[conda] torchaudio 0.13.0 py39_cu117 pytorch
[conda] torchvision 0.14.0 py39_cu117 pytorch

cc @ezyang @gchanan @zou3519 @seemethere @malfet @ngimel

@ZhiyuanChen
Copy link
Contributor Author

It was a clean conda install with no additional package, the command is
conda install pytorch torchvision torchaudio pytorch-cuda=11.7 -c pytorch -c nvidia

@samdow samdow added the module: binaries Anything related to official binaries that we release to users label Dec 12, 2022
@samdow
Copy link
Contributor

samdow commented Dec 12, 2022

Seems related to #89417

@bottler
Copy link
Contributor

bottler commented Dec 12, 2022

The workaround conda install libcusparse=11.7.3.50 -c nvidia given in #90673 might help.

@malfet
Copy link
Contributor

malfet commented Dec 12, 2022

Likely fixed by pytorch/builder#1227
@atalman can you please check

@atalman
Copy link
Contributor

atalman commented Dec 12, 2022

@malfet testing the fix now

@malfet
Copy link
Contributor

malfet commented Dec 12, 2022

Fixed, closing (please reopen/file new one if this is not the case)

@malfet malfet closed this as completed Dec 12, 2022
@malfet malfet added module: regression It used to work, and now it doesn't module: cuda Related to torch.cuda, and CUDA support in general module: third_party labels Dec 12, 2022
@ZhiyuanChen
Copy link
Contributor Author

I tried a clean install again but seems still not working

@ZhiyuanChen
Copy link
Contributor Author

The workaround conda install libcusparse=11.7.3.50 -c nvidia given in #90673 might help.

this does help

@atalman
Copy link
Contributor

atalman commented Dec 13, 2022

@ZhiyuanChen it should be fixed I do see our packages are updated here: https://conda.anaconda.org/pytorch/noarch/

pytorch-cuda-11.7-h67b0de4_1.tar.bz2 | 3 KB | 2022-12-13 00:53:26 +0000

pytorch-cuda-11.7-h67b0de4_1.tar.bz2 3 KB 2022-12-13 00:53:26 +0000

@ZhiyuanChen
Copy link
Contributor Author

@ZhiyuanChen it should be fixed I do see our packages are updated here: https://conda.anaconda.org/pytorch/noarch/

pytorch-cuda-11.7-h67b0de4_1.tar.bz2 | 3 KB | 2022-12-13 00:53:26 +0000

pytorch-cuda-11.7-h67b0de4_1.tar.bz2 3 KB 2022-12-13 00:53:26 +0000

I think I'll try again later.... Clean install is really painful

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
high priority module: binaries Anything related to official binaries that we release to users module: cuda Related to torch.cuda, and CUDA support in general module: regression It used to work, and now it doesn't module: third_party triage review
Projects
None yet
Development

No branches or pull requests

5 participants